TWM606822U - Raw material price prediction system - Google Patents

Abstract

A raw material price prediction system includes at least one operating device and a servo host. The at least one operating device includes an input module adapted to input at least one price data. The server host communicates with the operating device and receives the price data. The server host includes a data collection module, an analysis module and an output module. The data collection module is suitable for collecting multiple market data from an external Internet. The analysis module includes a first algorithm and a second algorithm, and is suitable for calculating the price data and the market data with a first algorithm, and generating multiple result data. The output module is suitable for transmitting the result data to the operating device.

Description

Raw material price prediction system

A prediction system, especially a raw material price prediction system.

When the factory produces goods, it is bound to purchase different raw materials from the market, especially certain bulk production raw materials (such as PVC powder). These bulk raw materials usually arrive about two months after the purchase. However, the prices of these bulk raw materials will fluctuate due to market conditions. If the volatility changes too much when the goods arrive, it will affect the purchase cost.

In order to achieve reasonable raw material procurement costs, the industry usually makes raw material price forecasts. However, the current price forecasting methods mostly rely on the empirical judgments of the purchasers. The accuracy is limited, and the market prices cannot be effectively grasped, especially when the market emerges Such as financial turmoil) is more difficult to predict.

Therefore, how to solve the above problems is worth considering for those with ordinary knowledge in this field.

In order to solve the above problems, this creation provides a raw material price prediction system, which can input historical price data and produce a price prediction model, which can more accurately predict raw material price changes.

This creation provides a raw material price prediction system, including at least one operating device and a server host. The at least one operating device includes an input module adapted to input at least one price data. The server host communicates with the operating device and receives the price data. The server host includes a data collection module, an analysis module and an output module. The data collection module is suitable for collecting multiple market data from an external Internet material. The analysis module includes a first algorithm and a second algorithm, and is suitable for calculating the price data and the market data with a first algorithm, and generating multiple result data. The output module is suitable for transmitting the result data to the operating device.

In the above-mentioned raw material price prediction system, the input device is also suitable for inputting an algorithm instruction.

In the above-mentioned raw material price prediction system, the analysis module receives the algorithm instruction, and selects the first algorithm or the second algorithm according to the algorithm instruction.

In the above-mentioned raw material price forecasting system, the analysis module establishes a forecasting model in a time series.

In the aforementioned raw material price prediction system, the first algorithm is Prophet, and the second algorithm is ARIMA (Autoregressive Integrated Moving Average Model).

10: Internet

100: Raw material price prediction system

110: operating device

111: Input Module

120: Servo host

121: Data Collection Module

122: Analysis Module

123: output module

21, 22, 23, 24, 30: graph

31: Actual price line

32: Forecast price line

33: Lower limit of error

34: Upper limit of error

Figure 1 shows the structure of the raw material price prediction system of this creation.

Figure 2 shows a schematic diagram of the analysis curve.

Figure 3 shows a schematic diagram of data verification.

This creation provides a raw material price prediction system that collects historical price data and market data, and predicts the price trend of raw materials through artificial intelligence algorithms, which can be used for reference by the industry and formulate appropriate procurement strategies.

Please refer to Figure 1. Figure 1 shows the architecture of the raw material price prediction system created by this creation. The raw material price prediction system 100 of the present invention includes at least one operating device 110 and a server host 120.

The operating device 110 is a device operated by a user, and is an electronic device that can be connected to the Internet, and can be a personal computer or a smart device. In this embodiment, the operating device 110 includes an input module 111, and the input module 111 is adapted to input at least one price data. The price information is the price of raw materials, and includes historical prices. Therefore, the price data also includes a time data, and the time data includes data such as year, month, and day.

The server host 120 is the main computing component of the raw material price prediction system 100, and can be a single server for computing, or a multiple server for joint computing. In addition, the server host 120 is communicatively connected to the operating device 110 and receives price data from the operating device 110.

In one embodiment, the server host 120 includes a data collection module 121, an analysis module 122, and an output module 123. The data collection module 121 is adapted to collect multiple market data from an external Internet 10. The data collection module 121 is, for example, a crawler program that can automatically collect data from the Internet. The collected data are, for example, market dynamics and other data.

The analysis module 122 is adapted to calculate the price data and the market data using a first algorithm, and generate multiple result data. In addition, the algorithm used by the analysis module 122 is a time series algorithm. That is to say, the analysis module 122 obtains price data and market data respectively, and uses the price data and market data to build a model with the first algorithm, thereby predicting the possible trend of future raw material prices and generating the predicted result.

In an embodiment, the analysis module 122 further includes a second algorithm, and the analysis module 122 can receive an algorithm command from the operating device 110, and select the first algorithm or the second algorithm according to the algorithm command. Method to calculate. In other words, the analysis module 122 includes at least two algorithms, and the operator can select the algorithm to perform calculations to obtain different prediction results. In one embodiment, The first algorithm is, for example, Prophet, and the second algorithm is, for example, ARIMA (Autoregressive Integrated Moving Average Model).

Prophet is a process of predicting time series data based on an additive model (Additive regression model). The characteristic is that nonlinear trends coincide with annual, weekly, and daily seasonality and holiday effects. It is very suitable for analyzing historical data with strong seasonal effects. The Prophet model is defined as follows: y(t)=g(t)+s(t)+h(t)+ε _t

Among them, g(t) is a piecewise linear or logistic growth curve trend. By selecting change points from the data, Prophet automatically detects trend changes. g(t) is the seasonal component of each year using Fourier series modeling and the weekly seasonal component of using dummy variables. h(t) provides a list of important holidays for users. ε _t is the error.

The ARIMA model is a form of regression analysis. Among them, AR (Autoregressive, autoregressive): refers to using the same variable, for example, the previous periods of x, that is, x 1 to x t-1, to predict the performance of xt in this period, and assuming that they are linear, the model is defined as:

I (Integrated) refers to the original data of the difference time series to make it stable. The MA (Moving Average) model means that the sequence can be explained by giving different weights to the random terms of the same period and the past, defined as: x _t = μ + w _t + θ _t ω _{t -1} + θ ₂ ω _{t -2} +…+ θ _q w _tq

The standard notation of ARIMA model has p, d, q parameters to indicate the type of model used. Among them, p is the number of autoregressive terms; d is the number of differences; q is the number of moving average terms.

The output module 123 is adapted to transmit the result data to the operating device 110. That is, the output module 123 can convert the analysis result into result data, a file format that the operating device 110 can display. The operator can browse these prediction results through the operating device 110.

Please refer to Figure 2, which is a schematic diagram of the analysis curve. After the analysis module 122 obtains the price data, it converts it into a graph according to the time data in the price data, such as graph 21, which is the relationship between price and time, the horizontal axis is time (T), and the vertical axis is price (P).

Next, the analysis module 122 performs time splitting on the graph 21 and generates graphs with different meanings. For example, graph 22 shows the long-term trend of prices. Graph 23 shows price fluctuations in each season. The graph 24 is the residual part. Through the market data, the abnormal changes of market events (such as financial turmoil, tariff setting, etc.) can be found from the graph 24. After analyzing the graphs 22, 23, and 24, the curve model of the price prediction can be obtained, and the result data can be generated and provided to the operator to download and participate.

Please refer to FIG. 3, which is a schematic diagram of data verification. FIG. 3 is a schematic diagram of inputting the price data of PVC powder from July 2019 to January 2020 into the raw material price prediction system 100 for prediction. After inputting the price data into the raw material price prediction system 100, a graph 30 will be generated. The graph 30 includes the actual price line 31, the predicted price line 32, the error lower limit 33, and the error upper limit 34. It can be seen from FIG. 3 that the predicted price line 32 predicted by the raw material price prediction system 100 is very close to the actual price line 31, and the actual price line 31 also falls between the lower error limit 33 and the upper error limit 34.

Only in November 2019, the Chinese mainland cancelled the anti-dumping duty on PVC powder, which led to the increase in the price of PVC powder, which caused the actual price line 31 to deviate from the predicted price line 32 and exceeded the upper limit of error 34. Obviously, market news must still be taken into consideration in the forecasting model. Therefore, the raw material price prediction system 100 of the present creation is equipped with a data collection module 121, which collects market information from the Internet 10 and incorporates it into the prediction model to improve the accuracy of the prediction.

The raw material price prediction system 100 created by this creation can use historical price data and market information to predict the future trend of raw material prices more accurately through artificial intelligence calculations, so that the industry can better grasp the raw material prices and formulate more favorable procurement strategies for the business.

This creation is described above with examples, but it is not used to limit the scope of patent rights claimed by this creation. The scope of its patent protection shall be determined by the attached scope of patent application and its equivalent fields. Anyone with ordinary knowledge in the field, without departing from the spirit or scope of this patent, makes changes or modifications that are equivalent changes or designs completed under the spirit of this creation, and should be included in the following patent scope Inside.

10: Internet

100: Raw material price prediction system

110: operating device

111: Input Module

120: Servo host

121: Data Collection Module

122: Analysis Module

123: output module

Claims (7)

A raw material price prediction system includes: at least one operating device, including an input module, suitable for inputting at least one price data; and a server host, which is communicatively connected with the operating device and receives the price data, including: a data collection The module is suitable for collecting multiple market data from an external Internet; an analysis module includes a first algorithm and a second algorithm, and is suitable for calculating the price data and data with a first algorithm The market data generates multiple result data; and an output module adapted to send the result data to the operating device; wherein, the first algorithm is Prophet, and the second algorithm is ARIMA (Autoregressive Integrated Moving Average Model). The raw material price prediction system according to claim 1, wherein the input device is further adapted to input an algorithm instruction. The raw material price prediction system according to claim 2, wherein the analysis module receives the algorithm instruction, and selects the first algorithm or the second algorithm according to the algorithm instruction. The raw material price prediction system according to claim 1, wherein the analysis module establishes a prediction model in a time series. The raw material price prediction system according to claim 1, wherein the operating device is an electronic device that can be connected to the Internet. The raw material price prediction system according to claim 5, wherein the operating device is a personal computer or a smart device. The raw material price prediction system according to claim 1, wherein the server host is a joint operation of multiple servers.

Images